Upload folder using huggingface_hub

1_process_GM12878_data.ipynb

@@ -0,0 +1,599 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "3b10aa61-97b1-4a44-bc43-3b8d3a13269c",
+   "metadata": {},
+   "source": [
+    "# Process GM12878 ENCODE DNase data"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2e74579b-292a-4117-9b2c-5e6c507e18cd",
+   "metadata": {},
+   "source": [
+    "## Set up wandb"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "2fb5794c-676f-42fc-b996-86bd93cc6fdb",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "Tracking run with wandb version 0.19.7"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       "Run data is saved locally in <code>/code/github/gReLU-applications/VEP_benchmark/wandb/run-20250306_212453-pt0gn2y8</code>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       "Syncing run <strong><a href='https://wandb.ai/grelu/GM12878_dnase/runs/pt0gn2y8' target=\"_blank\">prep</a></strong> to <a href='https://wandb.ai/grelu/GM12878_dnase' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/developer-guide' target=\"_blank\">docs</a>)<br>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       " View project at <a href='https://wandb.ai/grelu/GM12878_dnase' target=\"_blank\">https://wandb.ai/grelu/GM12878_dnase</a>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       " View run at <a href='https://wandb.ai/grelu/GM12878_dnase/runs/pt0gn2y8' target=\"_blank\">https://wandb.ai/grelu/GM12878_dnase/runs/pt0gn2y8</a>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import wandb\n",
+    "import os\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "import torch\n",
+    "\n",
+    "import grelu.io.bed\n",
+    "import grelu.data.preprocess\n",
+    "import grelu.visualize\n",
+    "\n",
+    "%matplotlib inline\n",
+    "wandb.login(host='https://api.wandb.ai')\n",
+    "project_name='GM12878_dnase'\n",
+    "\n",
+    "run = wandb.init(\n",
+    "    entity='grelu', project=project_name, job_type='preprocessing', name='prep',\n",
+    "    settings=wandb.Settings(\n",
+    "        program_relpath='1_process_GM12878_data.ipynb',\n",
+    "        program_abspath='/code/github/gReLU-applications/VEP_benchmark/1_process_GM12878_data.ipynb')\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c165a6b4-b3f9-4d47-b1e7-b7b398799572",
+   "metadata": {},
+   "source": [
+    "## Load peaks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "c956787d-8605-4868-9ffb-1d21b7676625",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "peak_file = \"ENCFF588OCA.bed.gz\" # Downloaded from encode"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "82336873-59b8-416f-99ce-9ea352376aca",
+   "metadata": {},
+   "source": [
+    "## Merge peaks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "215cdf36-0f5d-4faa-b921-989cd90d6f9a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!zcat ENCFF588OCA.bed.gz | awk -v OFS='\\t' '{print $1,$2+$10-250,$2+$10+250}' | bedtools sort | bedtools merge | awk -v OFS='\\t' '{mid=int(($2+$3)/2); print $1,$2,$3,\".\",\".\",\".\",\".\",\".\",\".\",mid-$2}'  > ENCFF588OCA.summit.500bp.narrowPeak"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "532139a2-7d6d-430f-9791-25249c3771ad",
+   "metadata": {},
+   "source": [
+    "## Load merged peaks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "66cf0efa-a201-478b-a669-72ad697bcb75",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "peaks = grelu.io.bed.read_narrowpeak('ENCFF588OCA.summit.500bp.narrowPeak')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "6641d278-0d32-43c9-a9c5-b7e2b28de619",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "230421"
+      ]
+     },
+     "execution_count": 17,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "len(peaks)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ec1be6cf-411c-4886-a81d-27c76e0a2fbb",
+   "metadata": {},
+   "source": [
+    "## Extend to sequence length"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "3efa5401-cc29-4cf0-b034-6d6127d6e9e0",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "230421\n"
+     ]
+    },
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>chrom</th>\n",
+       "      <th>start</th>\n",
+       "      <th>end</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>230418</th>\n",
+       "      <td>chrY</td>\n",
+       "      <td>56838273</td>\n",
+       "      <td>56840387</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>230419</th>\n",
+       "      <td>chrY</td>\n",
+       "      <td>56839860</td>\n",
+       "      <td>56841974</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>230420</th>\n",
+       "      <td>chrY</td>\n",
+       "      <td>56849042</td>\n",
+       "      <td>56851156</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "       chrom     start       end\n",
+       "230418  chrY  56838273  56840387\n",
+       "230419  chrY  56839860  56841974\n",
+       "230420  chrY  56849042  56851156"
+      ]
+     },
+     "execution_count": 19,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "peaks = grelu.data.preprocess.extend_from_coord(\n",
+    "    peaks,\n",
+    "    seq_len=2114,\n",
+    "    center_col=\"summit\"\n",
+    ")\n",
+    "print(len(peaks))\n",
+    "peaks.tail(3)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1521458e-29fd-456f-bd26-1d00bf4e86d0",
+   "metadata": {},
+   "source": [
+    "## Filter peaks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "e8403d98-11a4-4291-9544-a0ae05bee5cd",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Keeping 223259 intervals\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Filter peaks outside autosomes\n",
+    "peaks = grelu.data.preprocess.filter_chromosomes(peaks, 'autosomes')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "bedcfc1e-95f3-4cdc-bbce-5bd4312ab87b",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/opt/conda/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Keeping 222151 intervals\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Filter peaks close to blacklist regions\n",
+    "peaks = grelu.data.preprocess.filter_blacklist(\n",
+    "    peaks,\n",
+    "    genome='hg38',\n",
+    "    window=50 # Remove peaks if they are within 50 bp of a blacklist region\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c2c92489-dce0-488e-84de-4bd134e538e1",
+   "metadata": {},
+   "source": [
+    "## Get GC matched negative intervals"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "410fabfd-5d7a-4b27-946e-6958048c0064",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Extracting matching intervals\n",
+      "Filtering blacklist\n",
+      "Keeping 212904 intervals\n"
+     ]
+    },
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>chrom</th>\n",
+       "      <th>start</th>\n",
+       "      <th>end</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>37166</th>\n",
+       "      <td>chr1</td>\n",
+       "      <td>803320</td>\n",
+       "      <td>805434</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>188961</th>\n",
+       "      <td>chr1</td>\n",
+       "      <td>820232</td>\n",
+       "      <td>822346</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>111360</th>\n",
+       "      <td>chr1</td>\n",
+       "      <td>822346</td>\n",
+       "      <td>824460</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "       chrom   start     end\n",
+       "37166   chr1  803320  805434\n",
+       "188961  chr1  820232  822346\n",
+       "111360  chr1  822346  824460"
+      ]
+     },
+     "execution_count": 23,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "negatives = grelu.data.preprocess.get_gc_matched_intervals(\n",
+    "    peaks,\n",
+    "    binwidth=0.05, # resolution of measuring GC content\n",
+    "    genome='hg38',\n",
+    "    chroms=\"autosomes\", # negative regions will also be chosen from autosomes\n",
+    "    blacklist='hg38', # negative regions overlapping the blacklist will be dropped\n",
+    ")\n",
+    "negatives.head(3)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "id": "51e70cfc-aeb8-4ba3-8b6f-3b07380a9149",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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"
+     },
+     "metadata": {
+      "image/png": {
+       "height": 300,
+       "width": 400
+      }
+     },
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "grelu.visualize.plot_gc_match(\n",
+    "    positives=peaks, negatives=negatives, binwidth=0.05, genome=\"hg38\"\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c88c89f7-caac-4bf1-91c3-48907d084ba0",
+   "metadata": {},
+   "source": [
+    "## Combine all intervals"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "id": "94cee75f-192d-4172-9e8e-d1bc015cb751",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(435055, 3)"
+      ]
+     },
+     "execution_count": 25,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "regions = pd.concat([peaks, negatives])\n",
+    "regions.shape"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "887a92ee-5630-485f-879f-8728d0f64d9d",
+   "metadata": {},
+   "source": [
+    "## Save"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "17725634-c841-42bd-b79c-8b82965fae53",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "regions.to_csv('intervals.bed', sep='\\t', header=False, index=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "id": "936014bd-ee30-43fd-becd-a234e964a346",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<Artifact dataset>"
+      ]
+     },
+     "execution_count": 27,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "artifact = wandb.Artifact('dataset', type='dataset')\n",
+    "artifact.add_file('intervals.bed')\n",
+    "run.log_artifact(artifact)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "id": "dd7853c8-b729-455b-b11c-5a02d1f882ff",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       " View run <strong style=\"color:#cdcd00\">prep</strong> at: <a href='https://wandb.ai/grelu/GM12878_dnase/runs/pt0gn2y8' target=\"_blank\">https://wandb.ai/grelu/GM12878_dnase/runs/pt0gn2y8</a><br> View project at: <a href='https://wandb.ai/grelu/GM12878_dnase' target=\"_blank\">https://wandb.ai/grelu/GM12878_dnase</a><br>Synced 6 W&B file(s), 0 media file(s), 2 artifact file(s) and 0 other file(s)"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       "Find logs at: <code>./wandb/run-20250306_212453-pt0gn2y8/logs</code>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "run.finish()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "bb5cb230-eb3c-41bd-95e6-f15e0125d4d2",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.10"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

README.md

@@ -1,3 +1,40 @@
----
-license: mit
----
+---
+license: mit
+task_categories:
+- tabular-regression
+tags:
+- biology
+- genomics
+pretty_name: "GM12878 DNase regression model"
+size_categories:
+- 100K<n<1M
+---
+
+# GM12878_dnase-data
+
+## Dataset Summary
+This dataset contains genomic intervals used to train a regression model on GM12878 DNase data, described in Lal et al. 2025 (https://www.nature.com/articles/s41592-025-02868-z). Genome coordinates correspond to the hg38 reference genome. 
+
+## Repository Content
+The repository includes one BED file and one Jupyter notebook:
+1. `intervals.bed`: Genomic intervals stored in BED format.
+2. `1_process_GM12878_data.ipynb`: Jupyter notebook containing the preprocessing steps used to generate the `intervals.bed` file.
+
+
+## Dataset Structure
+
+### 1. Intervals file (`intervals.bed`)
+BED format (tab-separated). There are three columns with no header. All intervals have length 2114 bp.
+- Chromosome name.
+- start position.
+- end position.
+
+## Usage
+
+```python
+from huggingface_hub import hf_hub_download
+import pandas as pd
+
+file_path = hf_hub_download(repo_id="Genentech/GM12878_dnase-data", filename="intervals.bed")
+df_human = pd.read_csv(file_path, sep='\t', header=None)
+```